Underwriter pays merchant	−P = −I1 − I2 − I3	P = I1 + I2 + I3
First installment	−I1	+I1
Underwriter refund Customer	+I1	−I1
Merchant pays back underwriter	−P = −I1 − I2 − I3	P = I1 + I2 + I3A

An alternative cash flow may look like this:


	Customer:

	underwriter	Merchant

Underwriter pays merchant	−P = −I1 − I2 − I3	P = I1 + I2 + I3
First installment	−I1	+I1
Underwriter refund Customer	+I1	−I1
Merchant pays back underwriter	I2 + I3	−I2 − I3

A conventional chargeback is the reversal of a credit card payment that comes directly from an issuer which is an entity, such as an issuing bank, that issues accounts and debit/credit cards directly to consumers. A capturer, such as a capturing bank, is an entity that obtains payments, typically from the issuer, for the merchant. For merchants, chargebacks can be a frustrating threat to livelihood. For customers, chargebacks represent a shield between the customer and dishonest merchants.

In the installment payment context, a chargeback is a reversal of payments by the consumer to the underwriter, again typically the reversal of a credit card payment, that comes directly from an issuer.

On the outside, chargebacks can appear very similar to traditional refunds, yet there is one very relevant difference: rather than contact the merchant for a refund, the consumer is asking the issuer to forcibly take money from the capturer. An investigation follows, and if the capturer feels the cardholder's request is valid, funds are removed from the underwriter's account and returned to the consumer; the consumer, on the other hand, is in no way obligated to return whatever was purchased. Mechanically, this may be realized through a request through the capturing bank which the removes the funds from the “underwriter's account”

Unlike the situation with the refund, with a chargeback (or some other installment default), the underwriter may be out of pocket for the amount paid to the merchant. In most cases, the underwriter assumes the risk of the purchase, so the merchant is under no obligation to reimburse the underwriter. Note that this is credit extension, and not a purchase transaction, the underwriter is effectively purchasing the future value of the debt for the price of the transaction. The following is an example of a cashflow where chargeback occurs after the initial installment:


	Customer:

	underwriter	Merchant

Underwriter pays merchant	−P = −I1 − I2 − I3	P = I1 + I2 + I3
First installment	−I1	+I1
Underwriter refund Customer	+I1	−I1

In an instalment case, there may be a refund period available where the customer can return the item to the merchant, and the merchant could send the money back to the underwriter. However, if the item is beyond a refund period, and the customer choses to do a chargeback on an instalment payment, then the underwriter would likely start collection actions against the customer, the merchant would likely not be involved. In this case, the agreement would be between the underwriter and the customer where the underwriter extends credit to the customer, and the underwriter pays out the merchant. The underwriter likely would need to conduct collections actions if the customer refuses to pay their instalment payments as the merchant is out of the picture by this point.

Chargebacks/defaults can occur for various seasons. Some of these reasons are the blame of the merchant, and include:

a) product not received—this occurs when the merchant does not ship the product or makes a shipping error;
b) Product unacceptable—this occurs when the merchant ships the product, but there is something wrong with the product.

Some of these reasons are attributable to the customer, and include:

a) Identity Fraud—this occurs when the person making the purchase is not who they say they are. In this case, when the cardholder discovers the purchase, they will instigate a chargeback.
b) Chargeback Fraud—this occurs when the person making the purchase illegitimately claims a chargeback on a legitimate purchase. For example, the person can claim the chargeback on a product that is completely acceptable, and then keep the product even though they have been reimbursed, thereby obtaining the product for free.

The option of offering installment payments (IP) at checkout on an e-commerce has been made based on:

Merchant—installment payments offered for purchases from a given merchant or not;
Customer—credit history, etc; IPs offered based on the customer's credit history.

It is observed that some products have an increased associated risk of fraud or chargeback. For example, if any person buys a hoverboard, without taking into account who the merchant is or who the customer is, there may be a higher statistical likelihood of fraud/chargeback than if any person buys a stove. It may be too risky to justify offering installment payments on certain products. Existing installment decisions are not product specific, but are the same for all products offered by a given merchant. As such, installment payments may initially be offered for purchases from a specific merchant deemed not to be selling risky products; when that merchant starts to sell risky products, the IP option is removed for all products sold by the merchant, including the non-risky ones.

Product Specific Component into IP Decision

In accordance with an embodiment of the invention, a product specific component is included in the IP decision. This will allow IP to be offered for some products of a merchant, while not being offered for other products of the merchant. Up to three variables input to IP decision are determined:

1) Merchant-Specific Variable: M

Merchant variable (e.g. M) is based on the merchant offering the product for sale. Individual merchants have their own creditworthiness based on the quality of their service and likelihood they will incur chargebacks or cooperate with the underwriter. This can be based on past chargeback history for the merchant. For example, if 50% of products from a given merchant received chargebacks, then IP should not be offered at all. The merchant variable represents a measure of risk, associated with the merchant, to the underwriter. The merchant variable may also be based on whether a given merchant would benefit from IP. The merchant variable may be modified over time to reflect evolving transaction history.

2) Product-Specific Variable: P

The product-specific variable (e.g. P) is associated with the specific product being purchased, and reflects a measure of whether IP should be offered for that product. This can be a binary yes/no variable, or a value within a range, for example. In some embodiments, products are sorted into a plurality of risk categories. Which specific products are eligible for IP may, for example, be determined based on historical chargeback fraud. For example, based on historical behaviour, it may be appropriate to allow t-shirts to be purchased with IP, but not hair extensions. In some embodiments, the product-specific variable is entered on a different timeline than the IP decision. The variable can be entered as products are added to the merchant's catalogue, and can be updated from time to time to reflect the outcome of transactions. In some embodiments, the product-specific variable P, is surfaced to the merchant when products are added to the inventory.

In some embodiments, for at least some products, in determining the product-specific variable P, historical data is combined from multiple merchants that sell the product or similar products, and a single value for P is determined that is used for the product for any merchant.

In some embodiments, for at least some products, in determining the product-specific variable P, historical data is considered only from a specific merchant that sells the product or similar products, and a merchant-specific value for P is determined that is used for the product offered by the specific merchant.

The product-specific variable is product specific, so if there are multiple items in a shopping cart, there are multiple respective product-specific variables.

In some embodiments, machine learning (ML) is used to perform product classification. An input to the machine learning is a product-specific training dataset that maps an initial set of products (and associated training data, including for example, label data, description, title, image) to a risk category code of some sort, or more generally to a value of the associated product specific variable P. Then, associated data for products not included in the initial set are input to a machine learning algorithm for classification.

Note that ML algorithms are not perfect, and there will be some potential for erroneous classifications, that result in IP being extended where it otherwise would not, or withheld where it otherwise would be. The amount of erroneous classifications can be monitored over time to determine whether the algorithm is producing results with acceptable risk.

3) Customer-Specific Variable: B

The customer-specific variable (e.g. B) is associated with the specific customer. In some embodiments, for a new customer, there is a signup process to collect information and authenticate the customer. It is important to be able to ensure that the installments are in the right person's name.

The customer-specific variable may be based on, for example one or more of:

- transaction history for the customer—a customer with repeated purchases with no chargebacks would be less risky than a customer with a history of chargebacks;
- locally maintained, or 3^rdparty credit information—this can be used to minimize identity fraud; note this can factor in a consideration of the amount of the purchase, and the credit available to the specific customer. In some embodiments, the credit available for IP purchases for a given purchaser is adjusted over time as a function of historical activity.
- customer—e-commerce system interactions—certain behaviors of customers when interacting with e-commerce systems can be tracked and correlated with chargeback. Examples include length of time browsing the product web page, checkout history, types of products bought/interested in—frequent high risk purchaser, outside general area purchases (e.g. someone with a history of interest in clothing buys kitchen appliances); number of visits to an item's page.

In some embodiments, machine learning is used to determine a metric based on the third category of customer-specific information above. Historical data that correlates information about checkout history, customer behaviour etc. with chargeback likelihood is used to train an AI engine, and then in determining the customer-specific variable B for a specific transaction, the AI engine is executed based on current values for such information.

Various parameters such as the examples above can be combined to generate a value for the customer specific variable. Alternatively, multiple customer specific variables may be determined on respective bases.

Make IP Decision

The IP platform automatically generates an IP decision in time to present to the customer in the checkout page. It is not practical or realistic at this stage to have any human involvement; convenience is paramount for online shopping and any delay may turn the customer away. In the most general sense, the IP decision is some function of P, and optionally also of M and B. For example, IP=yes/no=F(M,P,B). The output of the IP decision determines whether IP will be offered at checkout for the particular product.

In some embodiments, the function F generates a weighted sum of the inputs:
F=w₁M+w₂P+w₃B

where w₁, w₂, and w₃are configurable weights.

In some embodiments, one or more of M,P,B is a binary value that must be true for F to be true. For example, if P is binary (0 or 1) and must be true, F could be expressed as F=P×F₁(M,B).

In some embodiments, a combination of the above is used, such that one or more of the values must pass a binary test to be true for F to be true, and the result of the F is affected by their weighted sum. For example, F could be expressed as F=M×P×B×F₁(M,P,B).

In some embodiments, a merchant provided input is also factored into the IP decision. If a given merchant does not want IP offered at all, then this merchant provided input indicates this, and is used on making the IP decision.

In some embodiments, the function output may be classified with one or more threshold values that will affect the IP offer. One or more set points may affect the length of the term offered, the interest rate, and the leniency of the payments.

In some embodiments, F is configured such that if a value for B reflects a sufficiently low risk, then F=1 irrespective of the value for M and/or B. For example, if a specific customer has a very strong transaction history over multiple IP transactions, they might be offered for a product normally associated with high risk.

As noted above, the product-specific variable is product specific, so if there are multiple items in a shopping cart, there are multiple respective product-specific variables. In some embodiments, a respective interim IP decision is made for each product, and the overall IP decision is yes, only if the interim IP decision is yes for every product in the cart.

In some embodiments, a respective IP decision is made for each product, and the customer is offered IP for the products with “yes” as the respective IP decision, and are not offered IP for the products that have “no” as the respective IP decision.

In some embodiments, the IP decision can be based on other variables, such as a maximum amount of credit to be offered.

FIG. 3 illustrates a system for determining and delivering an installment payment decision to a customer, according to one embodiment. The system includes aninstallment payment apparatus300, along with interior and auxiliary components.

Theinstallment payment apparatus300 includes a product specificvariable determiner302. The product specificvariable determiner302 may be a program, process, algorithm, or machine learning engine stored on non-transitory memory in a computing device. The product specificvariable determiner302 may be run on a generic processor of a computer device or may be implemented by dedicated circuitry, such as an application specific integrated circuit (ASIC), a graphics processing unit (GPU), or a field programmable gate array (FPGA). Optionally, the product specificvariable determiner302 includes a product-specific machine learning (ML)engine303 that determines the product specific variable.

The product specificvariable determiner302 may access data related to the products of a merchant's store. Examples of data collected include product names, descriptions, manufacturers, and reviews. The product specificvariable determiner302 may also access and retrieve product data from other merchants, internal records of the e-commerce platform, and supplied third party information. In the case that the product-specificmachine learning engine303 is used in the implementation of the product specificvariable determiner302, the data used to “train” the engine may be provided by any one or more of the listed locations.

The product specificvariable determiner302 may, upon request, calculate or generate a value or recommendation based on the data provided. This value may be binary, an integer, a floating point number, or some other indicator.

Theinstallment payment apparatus300 also includes a merchant specificvariable determiner304. The merchant specificvariable determiner304 may be a program, process, algorithm, or machine learning engine stored on non-transitory memory in a computing device. The merchant specificvariable determiner304 may be run on a generic processor of a computer device or may be implemented by dedicated circuitry, such as an application specific integrated circuit (ASIC), a graphics processing unit (GPU), or a field programmable gate array (FPGA). Optionally, the merchant specificvariable determiner304 includes a merchant-specific machine learning (ML)engine305 that determines the merchant specific variable.

The merchant specificvariable determiner304 may access data related to a merchant on an e-commerce platform. Examples of data collected include merchant personal and credit information, merchant history on the e-commerce platform (e.g. sales, customer satisfaction/reviews, other stores owned), the history of chargebacks against the merchant, and the settings and preferences of the merchant. The merchant specificvariable determiner304 may also access and retrieve product data from other merchants, internal records of the e-commerce platform, and supplied third party information. In the case that the merchant specificmachine learning engine305 is used in the implementation of the merchant specificvariable determiner304, the data used to “train” the engine may be provided by any one or more of the listed locations.

The merchant specificvariable determiner304 may, upon request, calculate or generate a value or recommendation based on the data provided. This value may be binary, an integer, a floating point number, or some other indicator.

Theinstallment payment apparatus300 further includes a customer specificvariable determiner306. The customer specificvariable determiner306 may be a program, process, algorithm, or machine learning engine stored on non-transitory memory in a computing device. The customer specificvariable determiner306 may be run on a generic processor of a computer device or may be implemented by dedicated circuitry, such as an application specific integrated circuit (ASIC), a graphics processing unit (GPU), or a field programmable gate array (FPGA). Optionally, the customer specificvariable determiner306 includes a customer-specific machine learning (ML) engine307 that determines the customer specific variable.

The customer specificvariable determiner306 may access and retrieve data related to a customer on an e-commerce platform. Examples of data collected include customer personal and credit information, customer history on the e-commerce platform (e.g. purchase history, return history, reviews), and the history of chargebacks that a customer has initiated. The customer specificvariable determiner306 may also access and retrieve product data from other merchants, internal records of the e-commerce platform, and supplied third party information. In the case that the customer specific machine learning engine307 is used in the implementation of the customer specificvariable determiner306, the data used to “train” the engine may be provided by any one or more of the listed locations.

The customer specificvariable determiner306 may, upon request, calculate or generate a value or recommendation based on the data provided. This value may be binary, an integer, a floating point number, or some other indicator.

The system further optionally also interfaces with a third party credit/customer authentication308, to obtain data that is comprised of up to date customer credit information, such as scores, purchase history, and balances. This data is provided to the customer specific variable determiner as the majority of customer information utilized in producing a value. More generally, the third party credit/customer authentication308 may provide information necessary to make a decision on the customer specific variable.

Theinstallment payment apparatus300 further includes anIP decision generator310. TheIP decision generator310 may be a program, process, algorithm, or machine learning engine stored on non-transitory memory in a computing device. TheIP decision generator310 may be run on a generic processor of a computer device or may be implemented by dedicated circuitry, such as an application specific integrated circuit (ASIC), a graphics processing unit (GPU), or a field programmable gate array (FPGA).

The IP decision generator can communicate and receive values, data, or other indicators from all three of the product specificvariable determiner302, the merchant specificvariable determiner304, and the customer specificvariable determiner306. TheIP decision generator310 may also communicate with acustomer facing interface312 to relay IP decisions to be interpreted and displayed to a customer. Various examples of the calculation of an IP decision have been presented above, and these can be implemented in the IP decision generator.

Thecustomer facing interface312 allows a customer to interact with theinstallment payment apparatus300 and the greater e-commerce platform. The customer facing interface may be implemented as part of a website, application, or point of sale (POS) terminal; it provides a customer with information regarding products, payment, and merchant information, and may redirect a customer who wishes to complete a transaction to anIP gateway318 or anormal payment gateway320. An example graphical user interface (hereafter referred to as a GUI) that may be implemented to interact with a customer is depicted inFIG. 5, described below. The customer facing interface may also communicate new or updated product, merchant, and customer information to theinstallment payment apparatus300. Thecustomer facing interface312 may include acheckout user interface313.

In some embodiments, there is a customer facing interface configurator311 interposed between the IP decision generator and thecustomer facing interface312. The customer facing interface configurator311 configures thecustomer facing interface312 based on the IP decision, to include the IP option or not.

TheIP gateway318 is a customer payment gateway that is capable of fulfilling a transaction via installment payments. TheIP gateway318 is used when the IP option is offered to the customer (based on the IP decision), and selected by the customer. Purchase and card information may be transmitted from theIP gateway318 to thecustomer facing interface312 or theinstallment payment apparatus300. The purchase processing of theIP gateway318 may be provided by the merchant, the e-commerce platform, the underwriter, or another party.

The normal (e.g. non-IP)payment gateway320 is a customer payment gateway capable of fulfilling a transaction in the conventional fashion without installment payments. Thenormal payment gateway320 is used when the IP option has not been offered to the customer (based on the IP decision), or when the normal payment gateway is selected by the customer. In this case, the checkout is processed to complete the payment. This may involve for example completing the transaction via a single credit or debit payment or via multiple such payments, all processed during checkout. Purchase and card information may be transmitted from the IP normal payments to thecustomer facing interface312 or theinstallment payment apparatus300. The purchase processing of thenormal payment gateway320 may be provided by the merchant, the e-commerce platform, the underwriter, or another party. In some embodiments, thenormal payment gateway320 andIP gateway318 are different and configured to handle transactions (or products in the same transaction) with different risk profiles. For example, thenormal payment gateway320 is configured to handle low or lower risk transactions/products (based on the IP decision) while theIP gateway318 is configured to handle high or higer-risk transactions or products.

The system further includes an underwriter configuration interface316, which allows an underwriter of the installment payments to configure the settings of theinstallment payment apparatus300 based on their preferences and the types of purchases, merchants, and products they wish to support with installment payment options. An example GUI that may be implemented to configure theinstallment payment apparatus300 is provided inFIG. 7 described below.

A merchant configuration interface314 may also be implemented as a part of the system. The merchant configuration interface314 allows a merchant to decide whether to allow IP offers to be made to customers on one or more of their products. An example GUI that may be implemented for merchant configuration is provided inFIG. 6, described below.

Should a customer try to make a purchase from a merchant that supports the offering of installment payments, a decision must be reached regarding whether IP is offered or not, and what the terms of the IP agreement are.FIG. 4 illustrates a method of making a decision in greater detail, according to one implementation.

Upon a customer checking out and entering their personal and/or card information, the following method may be performed. Atstep702, theinstallment payment apparatus300 gathers relevant data from available sources, within the preferences and boundaries set by the underwriter. Specifically, the product specificvariable determiner302 retrieves product data on the items in the customer's order, the merchant specificvariable generator304 retrieves data relating to the merchant fulfilling the order, and the customer specificvariable determiner306 gathers information on the customer. Examples of the specific types of data have been listed previously. The process at each variable determiner may occur simultaneously in parallel or sequentially in series.

Atstep704, a resulting value is produced at each variable determiner, which may be a binary value, an integer, a floating point number, or some other indicator. These values are henceforth referred to as P, M, and B for each of the product specificvariable determiner302, the merchant specificvariable determiner304, and the customer specificvariable determiner306, as noted previously. Other values may also be passed on as additional information.

Atstep706, the values are received by theIP decision generator310, and an IP decision is generated. The IP decision generator utilizes some function (i.e. “F”) to determine the IP decision. F is a function of P, and optionally one or both of M and B. The function may be implemented in numerous ways, with examples described previously, such as a weighted sum, a binary test of one or more values, a comparison to threshold values, any combination of such functions. TheIP decision generator310 consults the merchant configured settings and the underwriter configured settings to ensure its decision falls within the defined parameters.

If the decision is compliant with all configured settings, it is presented to the customer atstep708. The customer facing interface configurator311 configures thecustomer facing interface312 to display the results of the IP decision. If favourable, the IP decision presented to the customer may include payment amounts, length of installment term, and interest rates. In some implementations, the IP decision may have multiple options with different amounts and terms. In this case, the customer may be permitted to choose their preferred installment payment plan If an IP decision in unfavourable (i.e. IP is decided to be unavailable), this information may be communicated to the customer (e.g. “this order is not eligible for installment payments”) or the customer may not receive any indication. In some embodiments, once the customer confirms their preferred installment plan and completes the checkout process, the platform is configured to route the transaction using one or both of thenormal payment gateway320 andIP gateway318.

Atstep710, the data used by theinstallment payment apparatus300 may be updated. Examples of recorded information include current customer credit and/or personal information, the results of the decision, the values calculated by the variable determiner, and potentially transaction/order information.

The various steps and interfaces in the system require input from the customer, the merchant, and the underwriter. Various user interfaces to receive such input are disclosed below, and may be implemented as part of the system.

Customer Facing Interface

312

FIG. 5 illustrates an example GUI that may be used as a customer facing interface, according to one implementation. The interface may include asearch bar502 which is capable of using keywords to locate information on the online store, app, or other interface which may include products, shipping information, or other information regarding the merchant or products offered. A selection ofsite headers504 may be implemented for convenient navigation of the site.

When an IP decision is reached, it may be displayed to the customer as a part of the checkout process. An exampleinstallment payment offer506 is displayed as part of the customer facing interface316. Prices, payment terms, and interest rates are provided as part of theinstallment payment offer506. Items of an order that are eligible for installments are listed and included in the installment offer, while ineligible products are listed separately and added onto the down payment for the installment offer. In some embodiments, the offer may be partially configured by the customer. For example, the payment terms (i.e. the repayment time) may be decided by the customer within the limits set by the IP decision provided by theIP decision generator310; the amount of payments will be recalculated as a result. Theinstallment payment offer506 includes two buttons which may be chosen by the customer (e.g. with a click, tap, or gesture). The “Pay in Installments” button routes the customer to theIP gateway318; the “Proceed to standard checkout” button routes the customer to thenormal payment gateway320.

In some embodiments, when the customer is offered the installment payment option, the customer is also allowed to select, through a user interface, how many installments to divide the purchase over.

Typically, all the payments, including the initial payment and subsequent payments use the same payment mechanism (e.g. credit card, bank card, etc.). However, in some embodiments, a customer is given the option of using a different payment mechanism from payment to payment upon confirmation that the multiple payment mechanisms are identified with the same customer.

According to one embodiment, the GUI includes anorder summary bar508, which lists the items to be purchased as well as their prices. Other sums, additions, or subtractions to the price may also be listed, such as subtotals, totals, taxes, shipping costs, and discounts.

Merchant Configuration Interface

312

FIG. 6 illustrates an example GUI that may be used as a merchant configuration interface, according to one implementation. The interface may include asearch bar602 which is capable of using keywords to locate information on the online store, app, or other interface which may include products, shipping information, or other information regarding the merchant or products offered. Asidebar menu604 is included in the GUI as a potential navigation tool to provide shortcuts to other pages (e.g. orders, product pages, financial information).An installmentpayment configuration menu606 is provided as a part of the GUI ,according to one embodiment; it consist of one or more tools such as empty fields, selection boxes, drop down selection menus, and/or slider selections. The installmentpayment configuration interface606 allows a merchant to decide whether to let installment payments be offered on their products and whether a combination of installments and immediate purchases can be used by a customer. Specific products can also have preferences assigned to them by a merchant. A “no” from the merchant overrides any IP decision, but a “yes” from the merchant is still subject to the IP decision made by theIP decision generator310.

Anotification bar608 is also included in the GUI. The notification bar may list for the merchant orders that are being made on installment payments as well as updates from the e-commerce platform and/or the underwriter regarding products that are available or unavailable for installment payments, policy updates, and refunded or cancelled IP orders that a merchant must respond to.

Underwriter Configuration Interface316

FIG. 7 illustrates an example GUI that may be used as a merchant configuration interface, according to one implementation. The interface may include asearch bar702 that may be used to locate information on the underwriter's management platform, such as statistics, finances, related companies/clients, and relevant applications and websites. The underwriter may or may not be the e-commerce platform. The interface further includes aheader menu704 that allows for easy access and navigation to important or frequently used pages or interfaces of the management platform.

An installmentpayment configuration menu706 is included as a part of the GUI. The installment payment configuration menu allows an underwriter to control certain aspects or functionality of theinstallment payment apparatus300, namely, any one or more of the product specificvariable determiner302, the merchant specificvariable determiner304, and the customer specific variable determiner. Some examples of settings that could be changed or adjusted include risk thresholds, data to be included, the weighting of the measurements/values in determining the final decision, and specific permissions or overrides available to the underwriter. The settings may be segregated into basic and more specific advanced features as shown. An underwriter may choose to save or discard their changes by selecting either the “Apply Changes” or “Cancel Changes”, respectively. In another implementation, default settings may be available.

The GUI further includes anotification bar708. Thenotification bar708 may provide easy to access data regarding large data updates, merchant information, updates to theinstallment payment apparatus300, and other alerts to the underwriter.

Although the foregoing has been described with reference to certain specific embodiments, various modifications thereof will be apparent to those skilled in the art without departing from the scope of the claims appended hereto.

Numerous modifications and variations of the present disclosure are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims, the disclosure may be practiced otherwise than as specifically described herein.